Ωtimes and Ωhelvetica fonts under development: step one · why times and helvetica? first of all...

ΩTimes and ΩHelvetica Fonts Under Development: Step One

Yannis HaralambousAtelier Fluxus Virus, 187, rue Nationale, F-59 800 Lille, [email protected]

John PlaiceDepartement d’informatique, Universite Laval, Ste-Foy (Quebec) Canada G1K [email protected]

The Truth Is Out There

—Chris CARTER, The X-Files (1993)

Introduction

ΩTimes and ΩHelvetica will be public domainvirtual Times- and Helvetica-like fonts based uponreal PostScript fonts, which we call “Glyph Con-tainers”. They will contain all necessary charactersfor typesetting efficiently (that is, with TEX quality)in all languages and systems using the Latin, Greek,Cyrillic, Arabic, Hebrew and Tifinagh alphabets andtheir derivatives. All Unicode characters will becovered, although the set of glyphs of our Ω fontswill not be limited to these; after all, our goal is high-quality typography, which requires more glyphs thanwould be required for mere information interchange.Other alphabets will follow (the obvious first

candidates are Coptic, Armenian and Georgian) aswell as mathematical symbols, dingbats, etc.

Why PostScript instead of METAFONT? META-FONT is the ultimate tool for font development.Extending Computer Modern fonts to the Unicodeencoding is still one of our goals. We have starteddeveloping a set of fonts we call “Unicode Com-puter Modern” fonts, using techniques such as Vir-tual METAFONT (virtual fonts are created directlyby METAFONT using the gftotxt utility for textoutput). Nevertheless, we realized that the taskof writing METAFONT code for some thousands ofcharacters (including code for typewriter style) is atremendous task, which will take several years.So we have decided to take a small break

from METAFONTing, and to develop in a limitedtime period PostScript fonts that will cover amaximum number of languages and will give theTEX community a good reason to switch to Ω.

Why Times and Helvetica? First of all because,after Computer Modern, they are the most widelyused fonts in the TEX community. Many journals

and publishers request that their texts be typesetin Times; Helvetica (especially the bold series) isoften used as a titling font. Like Computer Modern,Times is a very neutral font that can be used in awide range of documents, ranging from poetry totechnical documentation. . .It would surely be more fun to prepare a

Bembo- or Stempel Garamond-like font for the serifspart and a Gill Sans- or Univers-like one for thesans-serifs part; but these can hardly be used in thescientific/technical area, and that’s perhaps whereTEX (and hence potentially Ω) is used the most.

When will the Ω fonts be finished? Thedevelopment of ΩTimes and ΩHelvetica fonts isdivided into four steps:

1. Drawing of PostScript outlines and packagingof Glyph Container fonts.

2. Development of virtual code, based on the realfonts of step one.

3. Kerning of virtual fonts.

4. Development of LATEX code and Ω TranslationProcesses necessary for the use of these fonts.

For the time being (June 1996) we have donethe biggest part of step one, and this is what wepresent in this paper. We hope to have finished withsteps two, three and four before the next teTEX CD-ROM in December 1996.

We want your support! Please keep in mind:

The choice and shapes of glyphs presented inthis paper are only a first attempt. We needyour feedback to improve them, so that youcan use them efficiently.

In the tables we present only Times family andmedium series fonts (except in the case of Tifinagh,which is also presented in the Helvetica family).Up-to-date tables of the remaining fonts can beconsulted on our Ω WWW server

http://www.ens.fr/omega

126 TUGboat, 17, Number 2—Proceedings of the 1996 Annual Meeting


Figure 1: Italic-style letters with ogonek (Polishand Lithuanian)

You can also retrieve the PostScript code from thesame address, or from

ftp://ftp.ens.fr/pub/tex/yannis/omega

General remarks on the fonts

To prevent confusion, the word “font” in this sectionis meant in the sense of the PostScript Type 1 fontstructure; and not of TEX text or math fonts: thefonts we describe in this paper will never be useddirectly for typesetting. Their raison d’etre is toprovide glyphs for the virtual Unicode+TypographyΩ fonts which we will develop in steps two–four.Hence, there is no need to look in the tables fora ‘e’: this character will be assembled by the virtualfont, using the glyphs of letter ‘e’ and of the acuteaccent.The same stands for the letter ‘c’ with cedilla:

it can be assembled out of the two correspondingglyphs; however, this is not true for letters withogonek: the shape of the ogonek changes while itgets attached to the letter; that is why you findletters with ogonek in the Glyph Containers andnot letters with cedilla. In Fig. 1 the reader cansee examples of letters in italic style, carrying anogonek accent.In Fig. 2 the reader can find the general

structure of the fonts:1 On the left, the 16-bitUnicode+Typography virtual font, on the right acertain number of Glyph Containers, that is 8-bitPostScript fonts.The reader will notice that a certain number of

glyphs are repeated in the different Glyph Contain-ers. This is because we want to minimize the numberof Glyph Containers used for a single-alphabet text.For example, accents for all fonts are stored in Glyph

1 In the figure, the reader will notice real font “AdobeZapf Dingbats”. In fact, the glyphs of this font have becomeUnicode characters (0x2701–0x27be) and we see no reason toredraw them since this font is widely available. Hence thevirtual 16-bit font will also point to the standard Adobe ZapfDingbats font.

Container “Common”; theoretically, to produce anacute-accented Latin letter and an acute-accentedCyrillic letter, one would use three Glyph Contain-ers: one for the accent, and one for each alphabet.To avoid this, we store all accents relevant to a givenalphabet, in the alphabet’s Glyph Container. Thesame method is used for shapes that are similar inthe different alphabets: Latin, Greek and Cyrillic al-phabets share the letter ‘A’, Latin, IPA and Cyrillicalphabets share the letter ‘a’.2

The “Common” Glyph Container

The “Common” Glyph Container, shown in Table 1,contains glyphs that will be used potentially inconjunction with all alphabets. These are describedin the following subsections.

Punctuation, digits, editorial marks Specialcare has been taken to distinguish between “typo-graphical” punctuation and “typewriter/computerterminal-derived” one: compare the typographicaldouble quotes lm and the straight ‘ASCII’ ones ".This table covers all Unicode punctuation

marks from the ASCII and ISO 8859-1 tables aswell as from the general punctuation table(0x2010–0x2046). We have not included a fewpunctuation marks specific to a single alphabet:Arabic asterisk, inverted comma and semicolon,Hebrew colon, Greek upper dot. These will be foundin the corresponding Glyph Containers.In Fig. 3 the reader can see how regular

curly braces have been transformed into squarebrackets with quill, by simply reflecting thecentral part of the brace.

Commonly adopted Latin alphabet derived

symbols Symbols like A use Latin alphabet lettersbut are used in many non-Latin-alphabet basedlanguages. Symbol z is an even stranger example:although the glyph ‘N’ does not exist in Cyrillic,this symbol is used mainly in Cyrillic-alphabetlanguages.

2 We will use a totally different approach when dvipsand Adobe Acrobat are able to use 16-bit PostScript fonts.Instead of having many ‘small’ PostScript fonts and one‘big’ virtual font, we will use a single ‘big’ PostScript font,in Unicode+Typography encoding. In that font, accentedletters and repeated identical shapes will be obtained by thePostScript font technique of composite characters. This willallow Acrobat users to select and copy Unicode-encoded textdirectly from the document window.Since the two techniques (16-bit + 8-bit real, vs. 16-

bit composite real) will share the same .tfm metrics forcharacters, it will be possible to convert .dvi files from oneformat to the other, so that files obtained today by thefirst method will be “Acrobat-ready” later, whenever Acrobatswitches to Unicode (hopefully soon!).

TUGboat, 17, Number 2—Proceedings of the 1996 Annual Meeting 127

Yannis Haralambous and John Plaice

OmegaTimesCommon

OmegaTimesLatin

OmegaTimesIPA

OmegaTimesGreek

OmegaTimesCyrillic

OmegaTimesHebrew

OmegaTimesTifinagh

Adobe™ ZapfDingbats

…

ΩTimes

virtual font

(Unicode

encoding +

characters

needed for

typography)

PostScript fonts

16-bit

8-bit

8-bit

8-bit

8-bit

8-bit

8-bit

8-bit

8-bit

8-bit

OmegaTimesArabicOne, …

Figure 2: General structure of the ΩTimes fonts (idem for ΩHelvetica)

Figure 3: The left and right squarebrackets with quill were drawn by reflectingthe central part of regular curly braces

In Fig. 4 the reader can see our small tributeto the GNU foundation: the “copyleft” symbol. Wehope that this character will soon be included in theletter-like symbols section of Unicode.

—Q. Why not use the \reflectbox macro toreflect the “copyright” glyph into a “copyleft” one?

—A. Because we want to treat “copyleft” as aseparate character in the virtual font, which maybe searched inside a .dvi or PDF file. In othercases, such as B, used in the A BBA logo and thecobar construction in Algebraic Topology, one can



Figure 4: The GNU “copyleft” symbol

Figure 5: The estimated symbol was drawninside a perfect circle

easily use PostScript-manipulation macros withoutdamage.Finally, in Fig. 5 the reader can see how the

estimated symbol fits inside a perfect circle (inred, for readers of a color version of this paper).

Currency symbols Among currencies havingproprietary symbols, there are strong and weakerones. The strong ones have made it into ISO 8859-1 (you-know-who made it even into ASCII itself,and is used by a well-known typesetting language

Letter ‘O’Letter ‘C’

Letter ‘e’

C

e

Figure 6: How shapes ‘C’, ‘O’ and ‘e’ were usedfor the design of the latin and cyrillic capitalletter shwa

to enter math mode. . . ), the remaining ones havefound their home in the currency symbols sectionof Unicode.We have included them all (even the Thai

currency symbol è, which looks suspiciously Latin)in the “Common” Glyph Container. Note that thesymbol Ñ for French Franc is virtually unknown inFrance. . .

Diacritics The zone 0xa0–0xe2 of the “Common”Glyph Container is dedicated to combining diacrit-ics. These diacritics are supposed to be usefulfor more than one alphabet; whenever a diacriticbelongs specifically to one alphabet, it has beenincluded only in the corresponding Glyph Container(this is the case, for example, of Vietnamese dou-ble accents, Greek spirit+accent combinations, andSlavonic accents). Thanks to the diacritics in the“Common” Glyph Container we will be able to con-struct all latin extended additional Unicodecharacters (0x1e00–0x1ef9) virtually, by combiningthem with letters from the “Latin” Glyph Con-tainer. This Unicode region covers Welsh, Viet-namese and transcriptions of Indic and other lan-guages.

The “Latin” Glyph Container

The “Latin” Glyph Container, shown in Table 2,contains glyphs of letters for Latin alphabet lan-guages. These are described in the following sub-sections.

Letters for Latin alphabet languages Allglyphs necessary to typeset Western and CentralEuropean, Nordic, Baltic, African languages, Viet-namese and Zhuang.Some African characters are derived from the

International Phonetic Alphabet. It is a fascinatingchallenge to design uppercase and italic-style formsfor these characters (for example, see in Fig. 6 the



Figure 7: The African letter f with hook inroman upper- and lowercase, as well as in italiclowercase form

Figure 8: Capital letter b with topbar(which shares the same glyph as cyrilliccapital letter be) small letter b withtopbar and cyrillic small letter be

uppercase version of letter © letter which derivesfrom the phonetic shwa, and in Fig. 7 the straightuppercase and straight lowercase versions of anAfrican letter that becomes a standard f in italicstyle).Sometimes, although a Greek form (or a form

derived from Greek) is used for the lowercase, theuppercase does not follow the Greek model: forexample, the uppercase of African letters ™ and ≠(the former is 100% Greek, while the latter looksmore like a “phonetic gamma”) are ä and ç.Inversely, sometimes a Greek form is used for

the uppercase only: ô is the uppercase form of theintegral-like π, a character derived from the IPA, forwhich one can hardly imagine an obvious uppercaseform—taking a Greek letter for that purpose is theeasiest solution.It is quite interesting how the lowercase of the

Latin letter Å differs from that of the Cyrillic lettersharing the same glyph (see Fig. 8).Special attention has been paid to the notorious

Dutch ligature ‘ij’, in italic Times lowercase formand in Helvetica uppercase form (see Fig. 9).

Figure 9: Italic letter ‘y’ with umlaut accentfollowed by the Dutch ligature ‘ij’ in Times italic,Helvetica medium uppercase and Helvetica bolduppercase form

Figure 10: The Vietnamese small letter owith horn was designed using the glyphs of thering accent, the apostrophe and letter ‘o’

In the former case, we have connected the twoletters, creating an intentional confusion with a ‘y’letter with umlaut accent. In Fig. 9 the readercan compare these two constructions. On the otherhand, the shape of the uppercase sans serif ‘IJ’ isderived from that of the letter ‘U’: this is commonDutch titling practice.The characters ^ and ~ are used in Breton.

They are not (yet) included in Unicode, and havebeen brought to our attention by Jacques Andre.Finally, in Fig. 10 the reader can see how the

“horn” of Vietnamese letters has been designed,using graphical elements from the font: in this case,the ring accent and the apostrophe.

Ligatures Besides the “standard five” ligatures Ã,Õ, Œ, œ, –, we have included ligatures for the casewhere:

• the second or third letter is an ‘i’ with ogonek:—, “, useful in Lithuanian;

• the second or third letter is a stroked ‘l’ : ”, ‘,useful in Polish;

• the second or third letter is a ‘j’: ’, ÷;

• instead of an ‘i’ one has an ‘ij’ ligature: ◊,ÿ,useful in Dutch;

• instead of ‘f’ one has a long ‘s’: Ÿ, ⁄, ¤, ‹,›. We haven’t (yet) included any ‘f’ + ‘long s’combinations.



Figure 11: French ligatures ‘st’ and ‘ct’ in Timesand Helvetica fonts

Finally we couldn’t resist the temptation ofmaking “French” ligatures fi, fl, of course, bothin Times and Helvetica (!) font families. Theseligatures are well-known because of their use (in theGaramond typeface) in the Pleiade book collection.One can argue about their reason for being inTimes (and especially in Helvetica) style, whichhas absolutely no historical background. . . Considerthem an experiment, and trust us for not makingthem automatic in default text mode!

Diacritics Accents 0xf8–0xff are used for Viet-namese only. Diacritics occupying positions 0xe8–0xf6 are shared with the “Common” Glyph Con-tainer.

The “IPA” Glyph Container

The “IPA” Glyph Container, shown in Table 3,contains a collection of glyphs needed for theInternational Phonetic Alphabet. They have beenfound in different sources: Unicode encoding,literature on phonetics (in particular we covered thecomplete table of characters of the French classicInitiation a la phonetique, by J.-M.-C. Thomas, L.Bouquiaux and F. Cloarec-Heiss, PUF, 1976). Onlya small number of these characters are contained inthe Unicode encoding. We would be grateful forany feedback from scholars on additional charactersor corrections of the existing ones.Since IPA is so. . . international, we have as-

sumed that one can encounter phonetic insertionsin text written in any language. Hence, we have in-cluded all glyphs needed, including lowercase Latinalphabet letters, and small capitals—the latter be-ing included only whenever these are significantlydifferent from their lowercase counterparts: includ-ing, for example, small caps ‘s’, ‘x’, ‘z’ would be

Figure 13: Three closely-related glyphs: German‘sz’, IPA ‘beta’ (not a Unicode character), Greeklowercase ‘beta’

Figure 14: Different types of ‘gamma’: the firstone from the Greek alphabet, the others from theIPA

useless since they are indistinguishable from lower-case ‘s’, ‘x’, ‘z’.This point deserves some explanation: one

should not confuse small capitals used for text andIPA small capitals. The former are a stylisticenhancement of text; they appear only in wordsentirely in small capitals style; their height is notequal to the x-height of the font, generally they areslightly higher. The latter are phonetic charactersused together with authentic lowercase letters: theymust have exactly the same height.In Fig. 12 the reader can see some small caps we

have designed for the ΩTimes IPA Glyph Container.It may not be obvious from the figure, but strokewidths of the small caps are exactly the same asthe ones of lowercase letters (see lowercase letter ‘a’to compare, as well as glyph ‘v’, which is used torepresent both a lowercase and a small caps ‘v’). Onthe second line of the figure, one can see the sameletters obtained as ‘fake’ small capitals.We had a lot of fun designing these characters.

In some cases, they just had to look a little differentfrom their Greek counterparts: for example, inFig. 13 the reader can see how the ‘phonetic beta’has been inspired by the German ‘sz’, and not by the‘real’ Greek beta; in Fig. 14 we present a collection



Figure 12: On the first line, lowercase letters ‘a’ and ‘v’ and specially designed IPA small capitals; on thesecond line, uppercase letters reduced 68%

Figure 15: Two IPA symbols of different origin:inverted letter ‘f’ and dotless ‘j’ with stroke

of gamma-like glyphs, as well as the ‘real’ Greekletter gamma.In some cases we were not very sure about

the origin of some IPA characters and made severalattempts: for example, in Fig. 15 one can see twosymbols with a superficial resemblance: an inverted‘f’ and a dotless stroked ‘j’. Which one is used inphonetics? The choice is left to the user.Finally there was one case where we had to solve

a real design problem: the one of the ‘l with retroflexhook’ and ‘ezh’ ligature (not a Unicode character).In most real-life examples we had the opportunityto see, the letter ‘ezh’ was sadly squeezed so that itstail remains higher than the retroflex hook of the ‘l’;we find it bad typographical practice to squeeze the‘ezh’ and propose three solutions (only one of which

Figure 16: Which one is the best ‘l with retroflexhook+ezh’ ligature? (Three proposals)

of course will survive in the final Glyph Container).In Fig. 16 the reader can see: (a) the tail of ‘ezh’merged with the retroflex hook of the ‘l’, (b) theretroflex hook of the ‘l’ continuing deep enough forit to be seen under the tail of ‘ezh’, (c) the retroflexhook rising higher so that it fits between the tail of‘ezh’ and the baseline. In cases (a) and (c) the riskmay be that the ‘l with retroflex hook’ be taken foran ‘ordinary l’ (compare ù, ¯ and ö); in case (b) weare going too deep under the baseline. . .Anyhow, we expect your feedback to resolve

this issue.

The “Greek” Glyph Container

The “Greek” Glyph Container, shown in Table 4,contains glyphs needed for the Greek language,ancient and modern. The problem with most



Figure 17: Two forms of Greek circumflex accent:the first used in modern Greek typesetting, thesecond one used in a number of scholarly typefaces

commercial Greek fonts is that they are either madefor modern Greek use, or for ancient Greek use bynon-Greek scholars. There is a third possibility:fonts made for ancient Greek use by Greek scholars.In this Glyph Container we have tried to satisfy

all three categories of users. Of course, this fontcan be used both for monotonic and polytonic text(there is a straight monotonic accent, while theacute one can be used as well). But we havegone even farther, by including two versions of thecircumflex accent, shown in Fig. 17: the tilde-likeone, used in Greece, and the cap-like one used inscholarly Western editions.3

Faithful to the first TUGboat paper by one ofthe authors (Haralambous and Thull, 1989), we haveincluded the inverted iota with circumflex accent,found in certain 19th-century editions of modernGreek (see Fig. 18).Version 1.0 of the Unicode standard contained

uppercase versions of Greek numerals, just as romannumerals were provided in upper- and lowercase;in ISO 10646 these characters were removed, ap-parently by decision of the Greek delegation. Wefind this action absurd (there are many examples ofuppercase numerals in literature) and have of courseincluded the relevant glyphs in the Glyph Container.

3 This brings us to a nice typographical joke: the Greek“circumflex” accent is either “tilde”-like or “cap”-like, butnever actually. . . “circumflex”-like!

Figure 18: The inverted ‘iota with circumflex’(not a Unicode character) used in 19th-centurymodern Greek printing

Figure 20: The different variant forms of Greekletters: beta, theta, phi, rho, kappa

In fact, we have included all known variants of Greeknumerals: stigma, digamma, qoppa and sampi (seeFig. 19, next page), and the upper and lower nu-meral signs.Greek letters are also very common in math-

ematics and physics. Some variant forms are usedfor different purposes in these fields (rho with curvedor straight tail, open/closed phi, open/closed theta,curly or straight kappa). To these we add a variantform used in regular text: the initial and medialbeta (b vs. 1). All variant forms of lowercase lettersare shown in Fig. 20.There is also a variant form of the letter sigma:

the so-called “lunate” sigma. This character is usedin some scholarly editions to avoid the distinctionbetween ordinary medial and final sigmas. In Fig. 21we compare it with Latin letter ‘c’: the lowercase oflunate sigma has no bulb and the upper one, no serif



Figure 19: Collection of glyphs used for Greek numerals; on the first line: lowercase stigma, digamma (2variants), qoppa (2 variants), sampi; on the second line: uppercase stigma, digamma, qoppa, sampi

Figure 21: The glyphs of Greek lunate sigma andLatin letter ‘c’ (in upper- and lowercase)

(and hence the Greek and Latin letters are identicalin the Helvetica family).The reader may ask why on positions 0x80–

0x8d and 0xa0–0xad of the Glyph Container tablewe have accented letters, while on positions 0x90–0x9d there are only accents and no letters. Theanswer is very simple: accents on Greek letterssometimes change shape according to the width ofthe letter. Greek vowels can be roughly dividedinto three width classes: narrow ones (the iota),wide ones (the omega) and medium ones (all theremaining). On row 8 and a we have placed accentsfor narrow and wide letters, respectively; on row9 we have placed the ordinary accents. And sincerows 8 and a have been made for unique vowels,we have prefered to include complete accent+lettercombinations (letters being aliases, of course) sothat the virtual font doesn’t need to make theconstruction.

The same reason justifies positions 0x5e and0x5f: the dieresis (dialytika, in Greek) does nothave the same width, depending on whether it isplaced on an iota or an upsilon.Finally, in the table there are also the specifi-

cally Greek guillemets (round ones), the upper dot,and the two forms of “subscript” iota: for low-ercase letters (ypogegrammeni) and for uppercaseones (prosgegrammeni).

The “Cyrillic” Glyph Container

The “Cyrillic” Glyph Container, shown in Table 5,contains glyphs needed for all languages usingthe Cyrillic alphabet, whether European or Asian.Characters for pre-Lenin Russian (fita, izhitsa, yat)have also been included, as well as “modernized”versions of Slavonic characters. As in the case of theLatin ‘st’, ‘ct’ ligatures, one may argue the necessityof modernizing Slavonic script, especially when itcomes to drawing the Helvetica version. It happensthat these characters have been included in Unicode(like the Coptic ones), and this is already reasonenough to draw the font; it is up to the user toactually adopt the “modern” font, or use beautifultraditional Slavonic typefaces.What was fascinating when designing Slavonic

letters was their relation to Greek ones. In Fig. 22we compare Slavonic and Greek Xi and Psi: ata first glance, Slavonic Xi bears a resemblanceto Greek lowercase xi, it is quite surprising that



Figure 22: Comparing Slavonic letters Xi and Psiwith the corresponding Greek ones

the Slavonic xi is reflected with respect to theGreek one. Uppercase Psi letters are identical,and lowercase Slavonic psi has a heavier stem(with serif, while Greek lowercase letters never haveserifs). Uppercase Slavonic Omega + is a magnifiedlowercase omega (where we have placed a serif onthe central stem). For the lower part of the SlavonicYus / we have used an inverted Psi: this is a smalldesigner’s secret allowing better integration of theletter in the Times (resp. Helvetica) font family.Another design initiative of ours was to use

graphic elements from the Serbian letter ! 1, forthe Asian Cyrillic å ú, § ¥, Ø ø. We expect userfeedback to validate or deny this choice.

The “Arabic” Glyph Container

The “Arabic” Glyph Container, shown in Tables 6–8 contains the glyphs that are necessary to typesetin any language using the Arabic alphabet.4 Thedesign of these Ω Arabic fonts doesn’t actuallyhave much in common the Times and Helveticafamilies; in fact, we have used popular moderndesigns, which can be used both for technical andliterary text, and which allow easy readability insmall sizes. Fat and thin stroke width, ascenderheight and descender depth have all been calculatedwith respect to the corresponding parameters of theLatin/Greek/Cyrillic fonts.In Fig. 23 the reader can see the metrical

relationship between the Latin, Arabic and Hebrewfonts: we want these characters to fit with one other,

4 We have also included undotted versions of letters ba,fa, qaf, for typesetting of old manuscripts.

so that multilingual texts using the three scripts willproduce typographically acceptable results.Concerning diacritics we have included all

vowels, and combinations with shadda and hamza,as well as some special cases: madda, wesla, verticalfatha, vertical fatha + shadda, and other diacriticsused in Arabic spellings of African languages,Kurdish, Baluchi, Kashmiri, Uigur and Kazakh.5

Esthetic Ligatures We have included a very smallnumber of esthetic ligatures (fewer than 150): ba-like letters followed by a final noon-like, initial fa-like letters followed by a ya-like, an initial lam-meem ligature, and the llah ligature with andwithout vertical fatha + shadda. We are notconvinced that heavy ligaturing of these fonts, inthe manner of traditional Naskhi fonts, would beesthetically judicious. Nevertheless, we are open toany suggestion on possible ligatures that might beadded.

The “Tifinagh” Glyph Container

The “Tifinagh” Glyph Container, shown in Tables 9and 10, contains the glyphs needed for typesettingthe Tamazight (alias Berber) language. A completedescription of the Berber TEX system developed byHaralambous [2].The glyphs shown on the tables warrant some

explanation. Tifinagh script has always beenwritten in a non-serif style. On table 9 we showa “Helvetica” version of the script, in the sensethat everything has been done to bring these glyphscloser to the Latin/Cyrillic/Greek Helvetica types.This approach is quite safe, and— in all modesty—the result should not surprise any speaker of Berber.On the other hand, table 10 shows a 100%

experimental font! The main idea was to say:“What would Tifinagh letters look like today ifthey had followed the same evolution as Latinones?” See Fig. 24 for a closer comparisonof some Tifinagh letters in both Helvetica andTimes styles. Admittedly, the result seems weird,and a lot of corrections have to be made beforethese drawings actually can be called Tifinaghglyphs. . . Nevertheless, the Berber language isbeing typeset more and more in its traditionalscript, often together with other scripts. As aresult, the problem of homogenization with Westerntypographical traditions must be faced; the style ofthe Times typeface is one of the first challenges.

5 We have not included diacritics used for the Qur’an,as we believe that these should be printed using very specifictraditional typefaces; nevertheless, the diacritics provided aresufficient for excerpts from the Qur’an.



Figure 23: Capital height, x-height, baseline and descender depth of Latin, Arabic and Hebrew letters

Figure 24: On line 1: Tifinagh letters in Helvetica style, with shapes identical to Greek and Latin ones;on line 2: idem. but shapes become stranger; on lines 3 and 4: the same letters, in Times style, using“evolutional logic”

The “Hebrew” Glyph Container

The “Hebrew” Glyph Container, shown in Table 11,contains the glyphs needed for Hebrew, Yiddish(both in the Russian or American YIVO spelling)and Ladino. The font design is based on a—very popular in Israel—modern Hebrew typeface.Once again we have adapted the stroke widthsand character dimensions of Latin/Greek/Cyrillicglyphs. In Fig. 23, the reader can compare the size

and weight of Hebrew letters with those of Latin andArabic.6

We have not included Masoretic signs in thefont, because we believe that the Bible should betypeset in traditional “square” fonts. Nevertheless,we have included letters with dagesh which appearonly in a few isolated positions in the Bible, as wellas the inverted nun. We have also included two

6 It should be noted that the height of Hebrew letters isexactly equal to the half distance between that of upper- andlowercase Latin letters.



different forms of the lamed-aleph ligature (with andwithout left stroke), used in older texts.

References

[1] Haralambous, Yannis and Klaus Thull. “Type-setting modern Greek with 128-character codes.”TUGboat 10,3 (1989), pages 354—359.

[2] Haralambous, Yannis. Un systeme TEX

berbere. Actes de la table ronde internationalepPhonologie et notation usuelle dans le domaineberbereq. Paris: Institut National des Langueset Civilisations Orientales, 1993. [Forthcomingin the Cahiers GUTenberg thematic issue onSemitic scripts.]



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * + , - . /

"3x 0 1 2 3 4 5 6 78 9 : ; < = > ?

"4x @ A B C D E F GH I J K L M N O

"5x P Q R S T U V WX Y Z [ \ ] ^ _

"6x ` a b c d e f gh i j k l m n o

"7x p q r s t u v wx y z | ~

"8x Ä Å Ç É Ñ Ö Ü áà â ä ã å ç é è

"9x ê ë

"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «» … À Ã Õ Œ œ

"Dx – — “ ” ‘ ’ ÷ ◊ÿ Ÿ ⁄ ¤ ‹ › fi fl

"Ex ‡ ·

"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 1: Tentative OmegaTimesCommon Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * + , - . /

"3x 1 2 3 4 5 6 78 9 : ; < = > ?


"5x P Q R S T U V WX Y Z [ \ ] ^




"9x ê ë í ì î ï ñ óò ô ö õ ú ù û ü

"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «» Ã Õ Œ œ

"Dx – — “ ” ‘ ’ ÷ ◊ÿ Ÿ ⁄ ¤ ‹ › fi fl

"Ex ‡ · ‚ „ ‰ Â Ê ÁË È Í Î Ï Ì Ó Ô

"Fx Ò Ú Û Ù ı ˆ ˜¯ ˘ ˙ ˚ ¸ ˝ ˛ ˇ"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 2: Tentative OmegaTimesLatin Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * + , - . /

"3x 0 1 2 3 4 5 6 78 9 : ; < = > ?


"5x P Q R S T U V WX Y Z [ \ ] ^ _





"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «» … À Ã Õ Œ œ

"Dx – — “ ” ‘ ’ ÷ ◊ÿ Ÿ ⁄ ¤ ‹ › fi fl

"Ex ‡ · ‚ „ ‰ Â Ê ÁË È Í Î Ï Ì Ó Ô

"Fx Ò Ú Û Ù ı ˆ ˜

¯"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 3: Tentative OmegaTimesIPA Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"3x 0 1 2 3 4 5 6 78 9 ; < = > ?

"4x A B C D E F GH I J K L M N O

"5x P Q R S T U V WX Y Z [ \ ] ^ _

"6x a b c d e f gh i j k l m n o



"9x ê ë í ì î ï ñ óò ô ö õ ú ù û

"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 4: Tentative OmegaTimesGreek Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * + , - . /

"3x 0 1 2 3 4 5 6 78 9 : ; < = > ?


"5x P Q R S T U V WX Y Z [ \ ] ^





"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «»

"Dx – — “ ” ‘ ’ ÷ ◊ÿ

"Ex ‡ · ‚ „ ‰ Â Ê ÁË È Í"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 5: Tentative OmegaTimesCyrillic Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '

"3x 0 1 2 3 4 5 6 78 9 > ?


"5x P Q R S T U V WX Y Z [ \ ] ^ _





"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «» … À Ã Õ Œ œ

"Dx – — “ ” ‘ ’ ÷ ◊ÿ Ÿ ⁄ ¤ ‹ › fi fl

"Ex ‡ · ‚ „ ‰ Â Ê ÁË È Í Î Ï Ì Ó Ô


Table 6: Tentative OmegaTimesArabicOne Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * + , - . /

"3x 0 1 2 3 4 5 6 78 9 : ; < = > ?


"5x P Q R S T U V WX Y Z [ \ ] ^ _





"Ax † ° ¢ £ § • ¶ ß® © ™ ´ ¨ ≠ Æ Ø

"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑∏ π ∫ ª º Ω æ ø

"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «» … À Ã Õ Œ œ

"Dx – — “ ” ‘ ’ ÷ ◊ÿ Ÿ ⁄ ¤ ‹ › fi fl

"Ex ‡ · ‚ „ ‰ Â Ê ÁË È Í Î Ï Ì Ó Ô


Table 7: Tentative OmegaTimesArabicTwo Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '( ) * +


"9x ê ë í ì î ï ñ óò ô ö õ ú"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 8: Tentative OmegaTimesArabicThree Glyph Container Table (June 10, 1996).

"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '

( ) * + , - . /

"3x 0 1 2 3 4 5 6 7

8 9 : ; < = > ?

"4x @ A B C D E F G

H I J K L M N O

"5x P Q R S T U V W

X Y Z [ \ ] ^ _

"6x ` a b c d e f g

h i j k l m"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 9: Tentative OmegaHelveticaTifinagh Glyph Container Table (June 10, 1996).



"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " # $ % & '

( ) * + , - . /

"3x 0 1 2 3 4 5 6 7

8 9 : ; < = > ?

"4x @ A B C D E F G

H I J K L M N O

"5x P Q R S T U V W

X Y Z [ \ ] ^ _

"6x ` a b c d e f g

h i j k l m"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 10: Tentative OmegaTimesTifinagh Glyph Container Table (June 10, 1996).

"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7

"2x ! " $ % & '( ) * + , - .

"3x 0 1 2 3 4 5 6 78 9 : < = > ?


"5x P Q R S T U V WX Y Z [ \


"7x p q r t v wx y z | ~"x8 "x9 "xA "xB "xC "xD "xE "xF

Table 11: Tentative OmegaTimesHebrew Glyph Container Table (June 10, 1996).


Ωtimes and Ωhelvetica fonts under development: step one · why times and helvetica? first of all...

Documents